Balancing innovation and safeguards for public legitimacy: A typology of sentiment toward AI regulation in the EU and US

The governance of Artificial Intelligence (AI) must balance innovation with safeguards to sustain public legitimacy. This study explores how participants in the European Union and the United States evaluate this trade-off by identifying archetypes of public expectations toward AI regulation. Drawing on survey data ( n = 292), we identify four archetypes of sentiment: Apprehensive Aspirationals (technologically literate skeptics favoring strong safeguards), Confident Champions (economically established and trusting AI optimists), Traditional Timids (older and less AI-literate skeptics), and Permissive Pragmatists (moderately enthusiastic about AI's benefits while opposing extensive government intervention). The prevalence of these archetypes differs across regions. While participants from the EU emphasize precaution and fairness, their US counterparts place greater weight on competence and performance. However, these patterns do not map neatly onto prevailing policy discourses or regulatory models. Instead, archetype membership is shaped more strongly by AI literacy, self-efficacy, and socio-economic position than by geography alone. Interpreted through a legitimacy lens, the archetypes reveal systematic combinations of input-oriented legitimacy concerns (procedural fairness, safeguards, and accountability) and output-oriented legitimacy expectations (effectiveness, competence, and innovation), rather than a simple trade-off between the two. The findings demonstrate that public legitimacy perceptions of AI regulation are configurational, integrating input and output legitimacy in patterned ways, and underscore the need for regulatory approaches that embed safeguards while enabling innovation to sustain legitimacy in AI governance. • Four archetypes of public sentiment toward AI regulation emerge from EU / US sample • EU residents favor precaution and fairness; US residents emphasize innovation and results • Two clusters mirror EU / US governance logics, two are independent of region • Perceived legitimacy of AI governance follows precautionary and performance dimensions